of February 12, 2018.This information is current as

(CR1/CD35)Intrinsic Affinity for Complement Receptor 1Complement Fragment C4b Have the Same The C4A and C4B Isotypic Forms of Human

Liliana Clemenza and David E. Isenman

http://www.jimmunol.org/content/172/3/1670doi: 10.4049/jimmunol.172.3.1670

2004; 172:1670-1680; ;J Immunol 

        average*  

4 weeks from acceptance to publicationSpeedy Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

?The JIWhy

Referenceshttp://www.jimmunol.org/content/172/3/1670.full#ref-list-1

, 21 of which you can access for free at: cites 53 articlesThis article

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2004 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on February 12, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

The C4A and C4B Isotypic Forms of Human ComplementFragment C4b Have the Same Intrinsic Affinity forComplement Receptor 1 (CR1/CD35)1

Liliana Clemenza*† and David E. Isenman2*

Several previous reports concluded that the C4b fragment of human C4A (C4Ab) binds with higher affinity to CR1 than doesC4Bb. Because the isotypic residues, 1101PCPVLD and 1101LSPVIH in C4A and C4B, respectively, are located within the C4dregion, one may have expected a direct binding contribution of C4d to the interaction with CR1. However, using surface plasmonresonance as our analytical tool, with soluble rCR1 immobilized on the biosensor chip, we failed to detect significant binding ofC4d of either isotype. By contrast, binding of C4c was readily detectable. C4A and C4B, purified from plasma lacking one of theisotypes, were C1�s converted to C4Ab and C4Bb. Spontaneously formed disulfide-linked dimers were separated from monomersand higher oligomers by sequential chromatographic steps. The binding sensorgrams of C4Ab and C4Bb monomers as analytesreached steady state plateaus, and these equilibrium data yielded essentially superimposable saturation curves that were well fitby a one-site binding model. Although a two-site model was required to fit the equilibrium-binding data for the dimeric forms ofC4b, once again there was little difference in the KD values obtained for each isotype. Independent verification of our surfaceplasmon resonance studies came from ELISA-based inhibition experiments in which monomers of C4Ab and C4Bb were equi-potent in inhibiting the binding of soluble CR1 to plate-bound C4b. Although divergent from previous reports, our results areconsistent with recent C4Ad structural data that raised serious doubts about there being a conformational basis for the previouslyreported isotypic differences in the C4b-CR1 interaction. The Journal of Immunology, 2004, 172: 1670–1680.

T he immune adherence receptor mediates attachment ofC3b- and C4b-opsonized particles to primate erythrocytesand phagocytic cells (1, 2). The protein responsible for

this activity, subsequently named complement receptor 1 (CR1),3

consists in its most common variant form of 30 extracellular re-peating modules, alternatively referred to as short consensus repeat(SCR) or complement control protein domains (3, 4). Based oninternal sequence homology, the N-terminal-most 28 domains maybe further subdivided into four long homologous repeats (LHRs),A, B, C, and D, each one consisting of seven SCR domains. Initialligand-mapping studies indicated that C4b binding was localizedto LHR-A, whereas LHR-B and LHR-C each had a binding site forC3b (5–7). It was further determined that most of the bindingenergy was contributed by the first three SCR domains of eachLHR, although the first four domains were required to yield bind-ing activity equivalent to the full-length LHR segment (7, 8). TheC4b binding site in SCR domains 1–3 became known as functionalsite 1, and the C3b binding sites in the nearly identical three do-

main combinations of SCRs 8 –10 and SCRs 15–17 of LHR-Band LHR-C, respectively, each became known as functional site2. It later became clear that site 2 was also capable of bindingC4b (9, 10), although there is dispute about whether the site 2interaction with C4b is weaker (9) or stronger (10) than that ofC4b with site 1.

In humans, C4 is encoded by two distinct genes,C4A andC4B. Although the resulting mature heterotrimeric proteinsconsisting of 93-kDa�-, 75-kDa �-, and 33-kDa�-chains are�99% identical with each other, the C4A and C4B isotypicvariants are defined by the presence of one of two hexapeptidesequences,1101PCPVLD1106 and1101LSPVIH1106 for C4A andC4B, respectively (mature human C4 numbering used through-out). These residues are located in the central C4d fragment(�42 kDa) of the�-chain�100 aa C-terminal to the thioester-forming residues, the latter mediating covalent attachment ofnascently activated C4b fragment to the C1-bearing target sur-face. Each of the isotypes displays allelic polymorphisms, withthe C4d portion of�-chain being the most polymorphic segmentof the molecule (11). Included in these are residues at fourpositions (1054, 1157, 1188, and 1191), which together withthose in the 1101–1106 isotypic sequences, give rise to theRodgers (Rg) and Chido (Ch) blood group alloantigens. Gen-erally, the Rg-determining polymorphic residues segregate withthe C4A isotypic residues, while Ch-determining residues seg-regate with the C4B isotypic residues, but there are rare excep-tions to this rule (12).

The most profound functional difference between the isotypes isin the nature of the covalent bond formed between the isotypic C4bfragments and the target surface, specifically an amide bond in thecase of C4Ab and an ester bond in the case of C4Bb (13, 14). Thisis in turn controlled by the residue at position 1106, with histidinedictating ester bond formation, whereas virtually any other amino

*Department of Biochemistry, University of Toronto, Toronto, Ontario, Canada; and†Department of Biopathology and Biomedical Methodologies, University of Palermo,Palermo, Italy

Received for publication July 11, 2003. Accepted for publication November 14, 2003.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby markedadvertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by Grant MOP-7081 from the Canadian Institutes ofHealth Research.2 Address correspondence and reprint requests to Dr. David E. Isenman, Departmentof Biochemistry, University of Toronto, 1 King’s College Circle, MSB 5306, Toronto,Ontario, Canada, M5S 1A8. E-mail address: d.isenman@utoronto.ca3 Abbreviations used in this paper: CR1, complement receptor type 1; C4BP, C4b-binding protein; Ch, Chido; fI, complement factor I; LHR, long homologous repeat;Rg, Rodgers; RU, resonance unit; SCR, short consensus repeat; sCR1, soluble CR1;SLE, systemic lupus erythematosus; SPR, surface plasmon resonance.

The Journal of Immunology

Copyright © 2004 by The American Association of Immunologists, Inc. 0022-1767/04/$02.00

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

acid will result in amide bond formation (15–17; mechanism re-viewed in Ref. 18). The nucleophilic preferences of the respectiveisotypic C4b fragments have been invoked to explain the greatertransacylation efficiency of C4Ab to amino group-rich immunecomplexes vs the greater transacylation efficiency of C4Bb to car-bohydrate-rich cellular target surfaces (14, 15, 19). The highertransacylation efficiency of C4Ab to immune complexes has inturn been used to explain the greater ability of native C4A, com-pared with C4B, to inhibit immune precipitate formation in serumor in serum-free mixtures containing Ag, IgG Ab, C1, and C4 (20,21). Other functional properties, including cleavage rate by C1�s,classical pathway C3 convertase subunit activity, regulation bycomplement factor I (fI) in the presence of C4b-binding protein(C4BP), and ability to act as a transacylation acceptor for C3b inthe formation of the classical pathway C5 convertase, were foundnot to differ substantially between C4A and C4B (13, 14, 22).However, there have been three reports concluding that C4Ab (or,in one case, ammonia-treated C4A, a thioester-cleaved C4b-likespecies) displayed higher binding to CR1 than did C4Bb, or am-monia-treated C4B (23–25). The study by Reilly and Mold (25)used isotypic forms of covalent dimers of human C4b fragment, inwhich the cross-link was via the liberated thioester cysteine, toactually measure equilibrium-binding constants. They determinedthat dimers of C4Ab displayed a �4-fold higher affinity for CR1than did dimers of C4Bb.

One reason for the interest in the functional properties of C4,and in particular of each isotype, is the apparent protective effectof C4 with respect to susceptibility to the autoimmune diseasesystemic lupus erythematosus (SLE; reviewed in Ref. 26). In hu-mans, there is a 75% prevalence of SLE with complete C4 defi-ciency, this being second only to C1q deficiency in which theprevalence is 93%. By contrast, there is no correlation of SLE withC3 deficiency, and the correlation with C2 deficiency is relativelyweak both in terms of frequency and severity of the symptoms.Although in humans complete deficiency states of C4 are quiterare, in healthy white individuals, homozygous C4A and C4B de-ficiencies occur with an estimated frequency of 4 and 1%, respec-tively (26). Although not seen in every SLE patient cohort (e.g.,Refs. 27 and 28), there have been multiple reports spanning twodecades of an increased frequency of complete, or even partial,C4A, but not C4B, deficiency states in SLE patient cohorts ascompared with healthy controls (29–37). The association of C4Anull alleles with increased risk for SLE transcended ethnic bound-aries, and thus was not readily attributable to the association of theC4A null allele to an extended MHC haplotype.

There are two nonmutually exclusive hypotheses for explainingthe strong association of early classical pathway component defi-ciencies and SLE. First, C4 may have a role in the induction ofself-tolerance against autoantigens (38, 39). Second, C4 has a rolein the disposal of immune complexes and apoptotic bodies, thelatter being loaded with precisely the autoantigens that one seesAbs against in SLE patients (26). The deposition of C4b fragmentonto the target both provides the subunit for classical pathway C3convertase formation and acts as a direct ligand for CR1, both ofthese events being important for the complement-facilitated dis-posal process. At present, nothing is known about the relativetransacylation efficiency of C4Ab vs C4Bb to apoptotic bodies.However, both the greater ability of C4Ab relative to C4Bb to bindcovalently to IgG, and thereby inhibit large and insoluble immunecomplexes from forming (20, 21), as well as the reported 3- to4-fold higher affinity of C4Ab for CR1 (25) can be envisaged aspreferentially facilitating the clearance of C4Ab-opsonized autoan-tigens. This, in turn, has been invoked as an explanation for theSLE association with C4A, but not C4B deficiency states.

Because the isotypic residues reside within the C4d region of themolecule, one might expect a contribution of this region to thebinding interaction with CR1 in a way that reflects the isotype-defining sequence differences. Specifically, the isotypic residuesmay either be involved in direct contact with the receptor, or theymay affect at a distance the conformation of a contact region inanother part of the C4d molecule. With respect to the latter pos-sibility, the presence of an additional proline (P1101) in the C4Aisotypic sequence appeared to be a promising candidate to mediatesuch a conformational effect. However, the recently determinedx-ray crystal structure of human C4Ad fragment did not supporteither one of these explanations (40). First, the loop containing theisotype-defining residues is in very close proximity to the site ofcovalent attachment, and as such, these residues are unlikely to beaccessible for direct contact by CR1. Second, although there cur-rently is no structure for C4Bd, based on the superimposability ofthe C4Ad structure on that of C3d, which is C4B-like in its isotypicsegment and thus lacks a proline in the first position, we havepreviously argued that in all likelihood its backbone conformationwill be indistinguishable from that of C4Ad (40). This would thenrender problematic the long-range conformational explanation forthe isotype-dependent differences in the binding of C4b to CR1.There remained the possibility that the CR1-binding differencesreflect contact differences with the major Ch/Rg epitopes definedby residues 1157, 1188, and 1191. As mentioned above, thesenormally segregate with the isotype-defining residues and, as vi-sualized in the C4Ad structure, they should be highly accessible(40). In view of the various questions raised by the structure ofC4Ad about the basis for reported CR1-binding differences ofC4Ab and C4Bb, as well as the possible relevance of this differ-ential affinity to our understanding of the etiology in humans ofSLE, we decided to re-examine this issue. Specifically, we haveused both surface plasmon resonance (SPR) and ELISA tech-niques to determine whether there is indeed an intrinsic affinitydifference between C4Ab and C4Bb for a soluble recombinantform of CR1 (sCR1). Additionally, we have assessed the re-spective abilities of isolated C4c and isotypic variants of C4dfragments to bind to sCR1.

Materials and MethodsPurified proteins

Human rsCR1 was a gift from Avant Immunotherapeutics (Needham,MA), and human C1�s was purchased from Advanced Research Technol-ogies (San Diego, CA). C4A and C4B were isolated, as previously de-scribed (41), from plasma lacking one of the two isotypes. Because theseproteins had been stored at �70°C for long periods of time, they containeda significant amount of thioester-hydrolyzed C4(H2O) material that wouldnot be cleavable by C1�s (42). Accordingly, these materials were repurifiedon a Mono-Q HR 10/10 FPLC column (Amersham-Pharmacia, BaieD’Urfe , Quebec, Canada), essentially as described by Hessing et al. (43),a procedure that is able to separate native C4 from C4(H2O). The purifiedproteins were converted to C4Ab and C4Bb, respectively, by overnightdigestion at 37°C with 1:100 w/w C1�s. The digestion mixture was thenchromatographed on a Mono-Q HR 10/10 FPLC column, as describedpreviously, as a means of separating monomer C4b from spontaneouslyformed disulfide-linked dimer C4b (43). Briefly, using a flow rate of 1ml/min and a starting buffer consisting of 50 mM NaCl, 20 mM Tris-HCl,and 2 mM EDTA, pH 7.4, the proteins were eluted from the Mono-Qcolumn with a 50 ml linear gradient to a final concentration of 0.5 M NaCl.C4b dimers and monomers were collected as separate fractions and thendialyzed against PBS (10 mM sodium phosphate, 0.15 M NaCl, pH 7.2).Following their respective concentration using Biomax Ultrafree-4 10-kDacutoff spin concentrators (Millipore, Bedford, MA), each fraction was sub-jected to gel filtration on an FPLC Superose 6 HR 10/30 column (Amer-sham-Pharmacia) equilibrated with PBS containing 0.02% NaN3 and run-ning at a flow rate of 0.3 ml/min. Before use in BIAcore experiments, theproteins were dialyzed into 10 mM HEPES, 75 mM NaCl, and 3 mM

1671The Journal of Immunology

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

EDTA, pH 7.4, and then surfactant P-20 (BIAcore, Piscataway, NJ) wasadded to a concentration of 0.01%.

C4c was obtained from purified pooled C4 (i.e., containing a mixture ofboth C4 isotypes) that was digested first with C1�s, as above, and subse-quently with factor I (1:30 w/w) and C4BP (1:50 w/w) for 18 h at 37°C.Factor I (44) and C4BP (45) were purified from human plasma, as de-scribed previously. C4c was then purified by the same two-step chromato-graphic procedure described above for C4b and exchanged by dialysis into10 mM HEPES, 75 mM NaCl, and 3 mM EDTA, pH 7.4 buffer, and forBIAcore experiments, surfactant P-20 was added to 0.01%.

Recombinant C4Adg and C4Bdg fragments, differing only with respectto their 1101–1106 isotype-specific residues, were constructed and ex-pressed in Escherichia coli, as previously described (40). Cell lysates wereloaded onto a 100-ml DEAE-Sephacel column (Amersham-Pharmacia)equilibrated with 10 mM Na phosphate, 25 mM NaCl, 2 mM EDTA, and0.1 mM DTT, pH 7.1 (buffer A). Following extensive washing, the C4dgprotein was eluted by increasing the NaCl concentration of the buffer to150 mM. DEAE-purified protein was dialyzed back into buffer A andloaded onto a Mono-Q HR 10/10 FPLC column equilibrated in this buffer.Elution was with a linear gradient to a final buffer containing 0.5 M NaClat a flow rate of 2 ml/min and a gradient duration of 20 min. The respectiveC4Adg and C4Bdg proteins were then further purified by gel filtration onan FPLC Superdex 200 HR10/30 column (Amersham-Pharmacia) equili-brated in 10 mM HEPES, 75 mM NaCl, and 3 mM EDTA, containing 0.1mM DTT.

The purity of all proteins was assessed by SDS-PAGE. The extinctioncoefficients (E280 nm

1% ) used were 8.2 for C4Ab/C4Bb and C4c and 13.6 forC4Adg/C4Bdg.

SPR measurements

The respective interactions of C4Ab and C4Bb with sCR1 were analyzedby SPR technology on a BIAcore X instrument (BIAcore). The experi-ments were performed at room temperature in a half physiologic ionicstrength buffer (10 mM HEPES, 75 mM NaCl, 3 mM EDTA, and 0.01%surfactant P-20, pH 7.4). sCR1 was diluted in 10 mM Na acetate, pH 5.5,at a concentration of 20 �g/ml and immobilized on flow cell 1 of a CM5chip (BIAcore) by using the amine-coupling kit from BIAcore. Flow cell2, to which mouse IgG (Sigma-Aldrich, St. Louis, MO) was immobilizedusing the same coupling conditions, was used as a control surface. A chipwith 6800 resonance units (RU) on the sCR1 ligand surface and 6350 RUon the control surface was used to test the binding of C4Ab, C4Bb, and C4cfragments for most of the experiments shown in this study. A chip with4900 RU on the sCR1 ligand surface and 4400 RU on the control surfacewas used to test the binding of the isotypic C4dg fragments. The various C4fragment protein analytes were used immediately after their last purifica-tion step. Binding was measured by injecting over the ligand-coupled chipseveral concentrations of analyte at a flow rate of 30 �l/min for 90–100 s.Following 180 s of buffer flow for dissociation, the sensor chip was re-generated with 1 M NaCl. The interaction of the C4dg fragments withsCR1 was also studied by reversing the immobilized ligand-analyte com-bination. Specifically, two different chips were generated by immobilizingC4Adg and C4Bdg to the respective sample flow cells (50 �g/ml in 10 mMNa acetate, pH 4.7, �1900 RU), and the control flow cells were in this casesimply sham activated and then ethanolamine deactivated.

Equilibrium phase-binding data were fit to one- and two-site models ofthe Langmuir binding isotherm using the nonlinear regression programMacCurveFit 1.5 (K. Raner Software, Victoria, Australia). The equationsdescribing these models were respectively:

�RU ��RUmax � �analyte�

KD � �analyte�(Equation 1)

�RU ��RUmax 1 � �analyte�

KD1 � �analyte��

�RUmax 2 � �analyte�

KD2 � �analyte�

(Equation 2)

where for Equation 2, the numerical subscripts refer to the equilibriumdissociation constant (KD) and maximal resonance units signal change atsaturation (�RUmax) associated with each class of binding site. The steadystate plateau-derived binding curves were also displayed as Scatchard

transformations having the following forms for the one-site (Equation 3)and two-site (Equation 4) models, respectively (46):

�RU

�analyte�� KA � �RUmax � KA � �RU (Equation 3)

�RU

�analyte�� 0.5 � �KA1��RUmax 1 � �RU � �KA2��RUmax 2 � �RU �

��KA1��RUmax 1 � �RU � KA2��RUmax 2 � �RU2 � 4KA1KA2�RUmax 1�RUmax 2

(Equation 4)

where for Equation 4, the numerical subscripts refer to the equilibriumassociation constant (KA) and maximal response units signal change atsaturation (�RUmax) associated with each class of binding site. Kinetic datawere analyzed globally using the BIAevaluation 3.0 software package. Thebinding model used for the C4b and C4c monomer sets of sensorgrams wasthe conformational change model: A�B7AB7AB* (analyte (A) binds toligand (B); complex AB changes to AB*, which cannot dissociate directlyto A�B). The binding model used for the C4b dimer curves was the par-allel binding model for a two-site heterogeneous binding model:A�B17AB1; A�B27AB2 (the binding curve obtained is the sum of thetwo independent reactions).

ELISA-based inhibition-binding assay

ELISA plates were coated with 100 �l of 10 �g/ml mixed-isotype C4bfragment. In a preliminary experiment in which variable concentrations ofsCR1 were added to the C4b-coated wells, we established that �50% ofmaximal binding occurred at a concentration of 1 �g/ml sCR1. For theinhibition experiments, this concentration of sCR1 was preincubated for1 h at room temperature with varying concentrations of C4Ab/C4Bb mono-mer (from 200 to 3.125 �g/ml) in a total volume of 100 �l. UncomplexedsCR1 in each preincubation mix was then allowed to bind to plate-boundC4b (1 h, room temperature), and its binding was detected by using theanti-CR1 mAb E11 (BD PharMingen, Mississauga, Ontario, Canada) as theprimary Ab and an anti-mouse IgG conjugated to alkaline phosphatase asthe secondary Ab (Jackson ImmunoResearch Laboratories, West Grove,PA). The assay was performed in 1/3 ionic strength PBS (10 mM Naphosphate, 50 mM NaCl, and 0.02% NaN3, pH 7.2).

ResultsSPR assessment of the interaction of the C4d and C4c fragmentswith CR1

Given that the C4d fragment forms an independently folded do-main entity (40) and the indications from the literature (23–25) thatthe C4d-localized isotypic segment in some way affects the inter-action of the parent C4b molecule with CR1, we first wished todetermine whether C4d on its own displayed any binding activityfor CR1, and if so, whether there were isotype-dependent differ-ences in binding activity. For these experiments, we have usedrC4d fragments corresponding in length to the physiologic fIcleavage-derived product (residues 938-1317), but, by analogywith the C3 fragment nomenclature, we have named them C4Adgand C4Bdg. This is to distinguish them from the N-terminally trun-cated C4d portion of the molecule that is visible in the x-ray struc-ture and that corresponds in length to the C3d proteolytic limitfragment of fI-derived C3dg. We tested several concentrations ofC4Adg or C4Bdg analyte (1–12 �M) for binding to biosensor-coupled sCR1, but found no binding signal above that seen on thecontrol IgG-coupled flow cell. Shown in Fig. 1 are representativeexperiments performed at 9.5 �M for each of C4Adg and C4Bdg.Because within each CR1 molecule there are potentially threebinding sites for C4b, and because the interaction of CR1 with anarray of C4b molecules on a target is an avidity situation, wewished to determine whether binding of sCR1 would be observedto an array of C4dg molecules, as this better mimics the physio-logic avidity situation. Accordingly, we reversed the analyte-li-gand combination from the first experiment and tested the bindingof several concentrations of sCR1 to biosensor-coupled C4Adg orC4Bdg. Once again, however, no specific binding was observed ineither case (results not shown).

1672 ANALYSIS OF CR1 BINDING TO C4Ab AND C4Bb

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

In view of the failure of isolated C4dg to bind to CR1, we nextasked whether the complementary fI-generated degradation frag-ment of C4b, namely C4c, could on its own bind to CR1. As canbe seen in Fig. 2, injection of variable concentrations of C4c ana-lyte (see Fig. 3 for SDS-PAGE analysis of the C4c material) overan sCR1-coupled flow cell resulted in a dose-dependent and spe-cific binding response. Because a steady state plateau was notreached for any of the concentrations of analyte tested, we wereunable to perform a direct analysis of the equilibrium-binding data.Global kinetic analysis of the binding of C4c to sensor-boundsCR1 using the BIAevaluation 3 software package did not yield asatisfactory fit to a simple 1:1 binding model. However, an accept-able fit (2 � 2) was obtained using a model in which a confor-mational change followed an initial binding event, and from thesekinetic parameters one could extract a value of 1.6 �M as an es-timated KD for the overall reaction (see Fig. 2 for further details).

SPR assessment of the interaction of isotypic variants of C4bmonomers and dimers with CR1

Although the isolated C4dg fragments do not have detectable bind-ing affinity for CR1 on their own, this region may still contributeto the binding of the parent C4b fragment, potentially in an iso-type-dependent manner. We therefore next analyzed the binding ofmonomeric and dimeric forms of C4Ab and C4Bb to biosensorchip-coupled sCR1. As reported previously by Hessing et al. (43)for mixed isotype human C4, we observed that activation of eitherC4A or C4B by C1�s resulted in 10–20% of the C4b material un-dergoing spontaneous disulfide-mediated dimerization. This mostlikely occurs via the cysteine residue liberated upon the thioesterhydrolysis that accompanies the proteolytic cleavage event. In ac-cordance with the published procedure, the monomer and disul-fide-linked dimer populations were separated by ion exchange onan FPLC Mono-Q column, and Fig. 4 shows the elution profile forC1�s-treated C4A. Also shown in this figure are the FPLC gel fil-tration profiles for each of the monomer and dimer peaks from theMono-Q column. The elution profiles for the C4B-derived materialwere very similar (data not shown). The SDS-PAGE analysis forthe purified monomer and dimer forms of C4Ab and C4Bb isshown in Fig. 3. It can be seen on the nonreduced gel that thedimer fractions contain relatively small amounts of contaminatingmonomer, and the level of contaminant is the same for both iso-types. The C4Bb monomer fraction appears to be totally devoid of

dimer and C4Ab contains only a trace amount (as well as a traceamount of C4c). The relative paucity of � -dimer band(s) uponreduction indicates that the vast majority of the dimers are disul-fide linked, as opposed to ester or amide linked, which would

FIGURE 2. Analysis of the interaction of C4c with sensor-bound sCR1.Sensorgram overlays generated by injecting several concentrations of C4c(indicated at the right of the curves in decreasing order from the conditionsof the uppermost curve) over a chip coupled with 6800 RU of sCR1. Here,and in subsequent BIAcore data panels, the curves shown represent the netbinding after the subtraction of the combined nonspecific binding and bulkrefractive index change on the control channel that had been coupled with6300 RU of human IgG. The interaction was analyzed by global kineticanalysis using the conformational change model in the BIAevaluation 3.0software package. The fitted curves are indicated with solid lines throughthe experimental time points. The conformational model is of the form:

A � BNka1

kd1

ABNka2

kd2

AB* (Equation 5)

where ka1 and kd1 are the forward and reverse rate constants for the en-counter complex and ka2 and kd2 are the forward and reverse rate con-stants for the conformational change. The overall equilibrium KD is givenby the formula (kd1/ka1) � (kd2/ka2) and was 1.6 �M in this case. Theindividual rate constant values are as follows: ka1 � 9.2 � 104 M�1 s�1,kd1 � 0.24s�1, ka2 � 6.8 � 10�3 s�1, kd2 � 4.3 � 10�3 s�1.

FIGURE 1. Lack of interaction of recombinant C4Adg and C4Bdg with sensor-bound sCR1. A concentration of 9.5 �M of analyte was injected for 90 sover a sensor chip. sCR1 was coupled to the experimental flow cell (4900 RU), and human IgG was coupled to the control flow cell (4400 RU). The C4dgcurves are shown as dashed lines; the control curves as dotted lines; and the net curves, obtained by subtracting the respective control RU signals from theexperimental RU curves, are shown in solid black. The arrows indicate the start and end points of the analyte injection.

1673The Journal of Immunology

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

represent thioester-mediated transacylation products. The materialdepicted on these gels is representative of the samples used in ourSPR measurements, these measurements being performed as soonas possible after the Superose 6 chromatography. We did, how-ever, note that upon storage at 4°C, dimer formation occurred inthe purified monomer fraction and that this appeared to be moreprevalent for the purified C4Ab monomers.

Depicted in Fig. 5A are the sensorgram overlay curves obtainedfor the interaction of the C4Ab and C4Bb monomer analytes in-jected over sensor-coupled sCR1 using an analyte concentrationrange of 0.06–1 �M. As can be seen readily for the first fiveincremental concentrations, in which the analyte concentrationswere perfectly matched, the two isotypes of C4b monomer be-haved in a very similar manner. Both displayed fast associationand fast dissociation kinetics and, as for C4c, the satisfactoryglobal fitting of the kinetic data required the use of a conforma-tional model. As the sensorgrams reached a steady state plateau inall cases, we were able to directly analyze the equilibrium data byboth nonlinear regression analysis of the saturation curves (Fig.5B) and Scatchard transformation of the data (Fig. 5C). Theseanalyses showed that the data for both C4Ab and C4Bb monomerswere well fit by a simple 1:1 Langmuir binding model and that thetwo saturation curves were essentially superimposable. The Scat-chard transformations similarly yielded nearly identical linearplots that were indicative of simple homogeneous binding. Al-though for both C4Ab and C4Bb the lowest concentration pointdeviated somewhat from linearity and may reflect a degree of bio-chemical heterogeneity in the binding reaction (see Discussion),the deviation could also be artifactual, as this part of a Scatchardplot is the most sensitive to deviations from ideal behavior as a

FIGURE 3. SDS-PAGE of the purified C4b and C4c fragments used forBIAcore analysis. The proteins were analyzed on an 8% gel under nonre-ducing conditions (top) and a 10% gel under reducing conditions (bottom).Molecular mass markers are indicated on the left.

FIGURE 4. Purification of monomers and dimers of C4b. C1�s-digestedC4A or C4B was first chromatographed on a Mono-Q column to separatemonomers from spontaneously formed dimers (top panel). The proteinswere eluted with a 50 ml NaCl gradient, as indicated. Monomer (middlepanel) and dimer (bottom panel) fractions were further purified on aSuperose 6 gel filtration column. The boundaries of the collected ma-terial for each peak are indicated. The data shown are for C4A, but weresimilar for C4B.

1674 ANALYSIS OF CR1 BINDING TO C4Ab AND C4Bb

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

result of small errors in the measurements. As can be seen in TableI, both isotypes of monomeric C4b displayed virtually identical KD

values of �0.9 �M, and the total binding capacity, as defined bythe respective �RUmax values, was similarly close.

A previous report that compared the binding of the isotypicforms of human C4b with red cell-associated CR1 used chemicallycross-linked dimers of C4b (bismaleimidohexane cross-linking viathe thioester cysteine) to take advantage of the avidity effect, andtherefore enhanced readout signal, that dimerization provides (25).Accordingly, we also measured the binding of our spontaneouslyformed disulfide-linked dimers of C4Ab and C4Bb to biosensor-coupled sCR1. As can be seen from the overlay plots shown in Fig.6A, the C4Ab and C4Bb dimers once again showed very similarconcentration-dependent binding behavior. This time, the best ki-netic fit was obtained with a parallel binding model to two inde-

pendent classes of binding site. Although we realize it to be some-what of an approximation, we have taken the �RU values at theend of the injection phase to represent pseudo-steady state plateauvalues, as this allowed a comparison of equilibrium phase func-tional affinity constants for the dimer forms of C4Ab and C4Bb forCR1. Consistent with the kinetic analysis, the minimal model thatcould fit the pseudo-steady state saturation data was that havingtwo classes of binding sites (Fig. 6B). This is most clearly indi-cated by the inverted hyperbola shape of the Scatchard-trans-formed and fitted data (Fig. 6C). Nevertheless, the values of theequilibrium dissociation constants extracted for each class of siteare very similar for the two C4 isotypes, as is their fractionalrepresentation of �RUmax (see Table I for a summary of all theequilibrium-binding data). Specifically, �13% of the binding sitesare of high affinity and have a KD of �10 nM, but the bulk of the

Table I. BIAcore equilibrium state analysis of the interaction of sCR1 with C4Ab and C4Bb

Monomers Dimers

1-site modela 2-site modelb

KD, �M �RUmax R2c KD1, �Md �RUmax1e (fraction RUmax) KD2, nMd �RUmax2

e (fraction RUmax) R2c

C4Ab 0.93 � 0.15 261 � 25 0.993 0.33 � 0.08 247 � 18 (0.86 � 0.06) 8.4 � 4.4 39 � 10 (0.14 � 0.03) 0.999C4Bb 0.94 � 0.14 281 � 23 0.990 0.39 � 0.15 291 � 39 (0.88 � 0.12) 11.7 � 7.7 39 � 15 (0.12 � 0.05) 0.999

a Parameters fit according to Equation 1 of Materials and Methods; error estimates of the fit are indicated.b Parameters fit according to Equation 2 of Materials and Methods; error estimates of the fit are indicated.c Correlation coefficient of the fit.d For the 2-site binding model, reported KD values are based on concentrations of the C4b covalent dimer species and would be 2-fold higher if calculated based on the

concentration of monomeric subunit.e Projected �RU change at saturation for each component of the binding, with the fractional numbers in parentheses being the individual component �RUmax divided by the

sum of these for the two components.

FIGURE 5. BIAcore analysis of the binding of purified monomers of C4Ab and C4Bb to biosensor-bound sCR1. A, Sensorgram overlay plots of thenet binding data as a function of analyte concentration, with these being indicated in descending order on each set of overlay curves. The sCR1-coupledchip used was the same as for the experiment displayed in Fig. 2. The solid lines through the experimental points represent the fitted data obtained by globalkinetic analysis using the conformational change model, as described in Fig. 2. The KD values determined from the kinetic analysis were 0.26 and 0.31�M for C4Ab and C4Bb monomers, respectively. B, Steady state analysis of the interaction of C4Ab and C4Bb monomers with sensor-bound sCR1. �RUvalues determined from the steady state plateau region were plotted as a function of analyte concentration, and the data were fit by nonlinear regressionaccording to the single site Langmuir binding model described by Equation 1 of Materials and Methods. The KD and RUmax values derived from these fitsare given in Table I. C, Scatchard transformation of the steady state binding data. The lines drawn represent insertion into Equation 3 of Materials andMethods the parameters obtained from the nonlinear fits shown in B.

1675The Journal of Immunology

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

binding is of lower affinity, having a KD of �0.4 �M. Because theanalyte concentrations were calculated on the basis of a dimermolecular mass of 380 kDa, if one calculates it on the basis of C4bsubunits present, this latter KD value becomes very similar in mag-nitude to what we had observed for monomeric C4b binding tobiosensor-coupled sCR1.

ELISA-based inhibition assay measuring the fluid-phaseinteraction of sCR1 with C4Ab and C4Bb monomers

As a means of verifying our SPR results via an independent ap-proach, we have used an ELISA-based inhibition assay in whichthe binding of sCR1 to plate-bound mixed-isotype C4b was eval-uated after preincubation with variable concentrations of mono-meric C4Ab or C4Bb as fluid-phase competitors. This assay formatwas chosen because any masking effect on C4b isotypic ligandpreference brought about by the covalent coupling of sCR1 to thebiosensor chip would not be a factor in this fluid-phase competi-tion assay. As shown in Fig. 7, and in accord with the SPR data,monomers of C4Ab and C4Bb were equipotent in their inhibitoryactivity.

DiscussionIn contrast to several previous reports in the literature (23–25), wefind that there is no difference in the intrinsic binding ability of theisotypic forms of human C4b fragment to human CR1. As will beelaborated upon below, the basic conflict between our results andthe earlier ones is not about the correctness of the earlier data, butrather that the higher binding to CR1 observed for C4Ab was beingascribed to an inherent property of its contact site for CR1. Instead

we will argue that the higher binding of C4Ab observed was mostlikely due to effects that conferred on it a functional affinity/avidityadvantage in a particular experimental system.

FIGURE 7. ELISA-based inhibition assay of the binding of sCR1 toplate-bound mixed isotype C4b by fluid-phase, monomeric C4Ab andC4Bb. The dashed line at the top of the figure indicates the absorbance inthe absence of fluid-phase inhibitors, and the background absorbance ob-tained in the absence of added sCR1, but with all other ELISA detectionreagents present, is indicated by the dotted line at the bottom of the figure.

FIGURE 6. BIAcore analysis of the binding of purified dimers of C4Ab and C4Bb to biosensor-bound sCR1. A, Sensorgram overlay plots of the netbinding data as a function of analyte concentration, with these being indicated in descending order on each set of overlay curves. The concentrations ofanalyte used are based on the dimer molecular mass of 380 kDa. The sCR1-coupled chip used was the same as for the experiment displayed in Fig. 2. Thesolid lines through the experimental points represent the fitted data obtained by global kinetic analysis using the parallel binding to two classes of bindingsite model. B, Steady state analysis of the interaction of C4Ab and C4Bb dimers with sensor-bound sCR1. �RU values derived from the end of the analyteinjection phase were plotted as a function of analyte concentration, and the data were fit by nonlinear regression according to the two-site Langmuir bindingmodel described by Equation 2 of Materials and Methods. The KD and fractional RUmax values derived from these fits are given in Table I. C, Scatchardtransformation of the steady state binding data. The lines drawn represent insertion into Equation 4 of Materials and Methods the parameters obtained fromthe nonlinear fits shown in B.

1676 ANALYSIS OF CR1 BINDING TO C4Ab AND C4Bb

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

Intrinsic affinity generally refers to the equilibrium-binding con-stant between a monovalent ligand and an acceptor molecule hav-ing a single class of binding site for that ligand. So long as theacceptor sites are sufficiently separate so that there is no stericinterference between them when ligand is bound, multivalency ofthe acceptor molecule will in principle not affect the observed af-finity of the monovalent ligand. Although to the best of our knowl-edge there is no definitive biophysical proof on this point, it hasbeen generally assumed that a single C4b molecule acts as a mono-valent ligand for CR1. CR1, however, has three potential bindingsites for C4b (9, 10), one in LHR-A (site 1) and one in each ofLHR-B and LHR-C (site 2). There is evidence in the literature thata dimeric form of C4(CH3NH2), a thioester-cleaved C4b-like mol-ecule, can bind to sCR1 in solution in a monogamously bivalentmanner, that is to two of the three potential C4b binding sites onthe same CR1 molecule (47). There is, however, dispute aboutwhether binding of C4b to site 1 has a relative affinity �3-foldhigher than that observed for site 2 (9) or whether the binding ofC4b to site 2 is, if anything, �2-fold stronger than that to site 1(10). In our experiments in which monomeric C4Ab and C4Bbwere injected over an sCR1-coupled biosensor chip, the data werewell fit to a model having a single class of binding site. If therewere significant contribution of a second class of binding sitewhose affinity differed by 3-fold or more from that of the first class,there should be more curvature in the Scatchard transformation ofthe data (Fig. 5C) than what was observed. Given that the reporteddifferences in binding strength in either direction of the two sitesare relatively small (9, 10), and also that for monovalent ligandthere is a 2:1 ratio of site 2 to site 1 present in full-length sCR1, webelieve that the most likely interpretation of the essentially homo-geneous nature of the binding data observed is that there are nomajor differences in affinity for C4b between sites 1 and 2. Thus,the observed KD of �0.9 �M for both C4Ab and C4Bb binding tointact sCR1 most likely reflects a weighted average affinity of col-lective binding to sites 1 and 2. However, because Reilly et al. (9)found that dimers of C4b bound to cells transfected with cDNAencoding full-length CR1 with the same affinity as to cells trans-fected with a cDNA encoding only site 1, and that in their handsthis affinity was 3-fold higher than to cells bearing only site 2, wecannot totally rule out the possibility that the homogeneous bind-ing that we observe is dominated by binding to site 1 only.

Because we were concerned that the surface effects inherent inthe SPR technology might be obscuring subtle differences in in-trinsic affinity of the isotypic forms of C4b for CR1, we also con-ducted an ELISA-based competition experiment that allowed one tomonitor the binding of monomeric C4Ab and C4Bb to sCR1 in so-lution. In this instance, too, however, the respective binding behaviorsof C4Ab and C4Bb to sCR1 were indistinguishable, thereby confirm-ing the results obtained using the BIAcore instrumentation.

It is worth mentioning that in choosing to work with monomericC4Ab and C4Bb so that we monitored intrinsic binding ability inboth the BIAcore and competition ELISA, the technical compro-mise that we had to make was to work at subphysiologic ionicstrength. In this way, it was possible to achieve a meaningful de-gree of binding site saturation for what is well established to be ahighly salt-sensitive interaction (10, 24). Although it is reasonableto have some concern about extrapolating our results to physio-logic ionic strength, we note that in one of the earlier studies thatreported higher sCR1 binding of a thioester-cleaved form of C4Arelative to C4B, the authors stated that the binding difference wasactually accentuated at low ionic strength (24).

Spontaneously formed disulfide-linked dimers of C4b, or evenC4(CH3NH2) (43, 47), have been reported previously, and pre-sumably these form via the thioester cysteine. As expected, dimers

of both C4Ab and C4Bb did show enhanced binding to biosensor-coupled sCR1 due to avidity effects, but in this instance too themagnitude of the effect was essentially the same for each C4 iso-type. The high affinity component was �100-fold greater than theaffinity observed with C4b monomers (Table I). This is commen-surate with the approximations made by others when they com-pared the relative ability of unlabeled monomers and dimers ofC4b, or C4(CH3NH2), to inhibit the binding of their labeleddimeric counterparts to the endogenous CR1 on human red cells(9, 47). What was somewhat unexpected was that higher aviditycomponent accounted for a relatively small fraction (�13%) of thetotal binding, whereas the majority of the binding had a KD com-mensurate with binding of C4b monomers (Table I). Either theprocedure used to couple sCR1 to the biosensor chip had madebivalent attachment a sterically unfavorable event, or the timecourse of the BIAcore experiment (100 s) is too short for a moresubstantial amount of bivalent binding to occur. With respect to thelatter point, although it was done at a lower temperature than ourexperiment (0°C vs 22°C), it has previously been reported thatbinding of C4b dimers to red cell CR1 displayed very slow kinet-ics, taking 90 min to reach equilibrium (9).

Inferences drawn from the structure of C4Ad, and its compari-son with the structure of C3d (40), had essentially predicted whatwe have observed experimentally, namely that the isotypic se-quence differences should not influence the respective intrinsicCR1-binding affinities of C4Ab and C4Bb. The crux of the argu-ment was that the backbone segments in and around the thioester,as well as the three-dimensionally proximal isotype-defining se-quences, were completely superimposable. Yet, the C4A-specificsequence PCPVLD was at least as dissimilar from the correspond-ing C3d segment DAPVIH as it was from the C4B-specific iso-typic sequence LSPVIH. In particular, the extra proline of theC4A-specific sequence was without conformational consequence.Although there is no structure available for C4Bd, it is highlyunlikely to be different at the level of backbone structure from thatof C4Ad. Consequently, there is no conformational basis for theisotype-defining segment influencing a remote CR1 contact site inC4d. Furthermore, direct contact of the isotypic residues with CR1is probably precluded by their proximity to the covalent attach-ment site. We know from previous serotyping (48) that the C4Aand C4B proteins that we used were of the common A3 and B1allotypes, and that they had also been typed as being Rg�Ch� andRg�Ch�, respectively. Although the Ch/Rg epitope-determiningresidues at 1157, 1188, and 1191 are surface exposed in the C4dstructure, and are well away from the site of covalent attachment,the equivalence of the intrinsic affinities of C4Ab and C4Bb forCR1 that we find shows that these residues also do not directlyaffect this interaction.

We believe that the difference between our results and the threeprevious reports claiming that C4Ab displayed a higher affinity forCR1 than did C4Bb can be attributed to issues related to the char-acterization of the ligands. In the study by Gatenby et al. (23),radioiodinated IgG-containing Ag-Ab complexes were opsonizedin the presence of purified C1� with either C4A or C4B before theirbinding to CR1-bearing human red cells was determined. Althoughsubstantially higher binding of the radioiodinated immune com-plexes to CR1 was observed when C4A was the opsonin, the studymade no correction for what was certain to be a higher depositionof C4Ab, than C4Bb, to the same amount of immune complex.Based on the experiments of Kishore et al. (19), for the sameamount of IgG transacylation target and constant C1� , C4Atransacylates onto IgG H chain at a 3- to 4-fold higher efficiencythan does C4B. Additionally, because the transacylation target nu-cleophiles within the immune complex are going to be different for

1677The Journal of Immunology

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

nascent C4Ab and C4Bb, the nature of the ligand clusters formedcould affect the binding to CR1, which is itself clustered on redcells (49). Thus, the higher binding of C4Ab-opsonized complexesthat was observed by Gatenby et al. (23) is most likely accountedfor by the well-established covalent binding differences of the twohuman C4 isotypes.

The study by Reilly and Mold (25) compared the functionalaffinity of radiolabeled dimers of C4Ab and C4Bb to CR1 on hu-man red cells and reported a �4-fold higher affinity for dimers ofthe C4A isotype. These workers presumably chose to take advan-tage of the avidity properties of dimers because their binding as-say, which included three wash steps, would not have yielded asufficient binding signal with monomeric ligand. Reilly and Moldwere well aware that the presence of higher molecular mass oli-gomers would bias their results due to an enhancement of the avid-ity effect. Consequently, they put their bismaleimidohexane-cross-linked proteins through a size exclusion chromatography step,which by their estimation reduced the proportion of higher molec-ular mass aggregates to less than 3%. The gel that they show oftheir covalent C4Ab and C4Bb dimers was run under reducingconditions, which would not break the bismaleimidohexane cross-links between � -chains, but would reduce disulfide-linked oli-gomers. Based on our observations on the unique properties ofC4Ab in forming disulfide-linked oligomers beyond the dimerstage, we strongly suspect the presence of such species uniquelywithin the aggregate portion of their C4Ab dimer preparations, butthese would only have been detectable on a nonreducing gel. Spe-cifically, it has been our experience that C4Ab has a higher pro-pensity to form spontaneous disulfide-linked dimers than doesC4Bb. Additionally, on nonreduced, but not on reduced SDS-PAGE, the C4Ab dimer fraction from the Mono-Q column con-tains an additional higher molecular mass species that is not seenin the C4Bb dimer fraction (data not shown). Our interpretation ofthese observations is that for C4Ab, spontaneous disulfide-medi-ated oligomerization can occur not only through the thioester cys-teine (C991), but also through the isotypic region cysteine(C1102). This would not only kinetically facilitate more efficientdimerization, but would also allow for the formation of species thatare larger than dimers. The accessibility of C1102 to the solvent,and therefore availability for disulfide bond formation, is apparentfrom inspection of a surface representation of the C4Ad structure(Protein Data Bank file 1HZF). By contrast, cysteine-mediateddimerization of C4Bb would stop at the dimer stage because res-idue 1102 in this case is a serine.

The study of Gibb et al. (24) used ammonia treatment of C4Aand C4B to cleave the thioester bond and induce a C4b-like con-formation. When equal amounts of ammonia-treated C4A and C4Bwere coated onto ELISA wells, it was found that the binding of125I-labeled sCR1 was �2-fold greater to the C4A material than toC4B. Because the ammonia treatment would liberate the thioestercysteine, as well as the isotypic cysteine of C4A, not only wouldthere be spontaneous disulfide-linked dimers formed somewhatmore preferentially for C4A, but as discussed above, higher oli-gomers would only be possible in the case of C4A. Such a clusterof C4Ab-like molecules may be better suited for capturing thesCR1 by binding simultaneously to up to three sites on the sameCR1 molecule, thus conferring an advantage relative to the wellscoated with ammonia-treated C4B.

Although neither C4Adg nor C4Bdg on its own was able to bindto CR1, substantial binding was observed for C4c. This parallelsthe behavior of C3 degradation fragment binding to CR1, namelythat C3c on its own binds to CR1, albeit with weaker affinity thanC3b, whereas C3d shows no binding whatsoever to CR1 (50, 51).Because the sensorgrams of C4c binding to an sCR1-coupled chip

did not reach a steady state plateau, it was only possible to extractbinding affinity data though analysis of the kinetics. In our generalexperimental design, we were most interested in using the equi-librium plateau regions of the sensorgrams as a readout of thebinding C4Ab and C4Bb monomers, as this method of analysis isnot influenced by known kinetic artifacts of the BIAcore instru-mentation. These involve limitations on mass transport to analytebinding sites within the carboxylated dextran matrix, which leadsto an underestimation of kon values, and rebinding of ligand duringthe dissociation phase of the experiment, which leads to an under-estimation of koff values. Because for the equilibrium-phase anal-ysis one wishes to maximize signal change, the experiments wereconducted at relatively high RU values (6800) of sensor chip-cou-pled sCR1 and also using relatively high analyte concentrations sothat binding data extending through at least 60% of the saturationcurve could be achieved. Unfortunately, these are precisely theconditions that exacerbate the problems in kinetic measurements.Accordingly, we put little credence into the absolute value of thekinetic constants, or the equilibrium constants derived from them.Nevertheless, because C4c and the monomers of C4Ab and C4Bbwere analyzed on the same chip, and required the same confor-mational mechanism to kinetically fit the data, it may be valid tocompare in relative terms the respective kinetically derived equi-librium-binding constants. Whereas for C4Ab and C4Bb, the ki-netically derived KD values were respectively 0.26 and 0.31 �M(Fig. 5), that for C4c was 1.6 �M (Fig. 2), suggesting �5-foldweaker binding relative to the parent C4b ligand. Given the com-plete absence of CR1-binding activity of C4dg on its own, we inferthat the higher affinity of C4b for CR1 represents an indirect scaf-folding effect whereby one or more contacts between the C4d do-main and the larger C4c fragment may stabilize the conformationof CR1-contacting segments located within the C4c part of the C4bmolecule.

The fact that the kinetic data obtained with monomers of C4c,C4Ab, or C4Bb could not be fit by a simple 1:1 encounter model,but rather required invoking a kinetically coupled unimolecularconformational change subsequent to the initial binding event, mayalso be meaningful in terms of the nuclear magnetic resonancestructure of CR1 site 2 (SCR domains 15–17) and the mappingonto this structure the location of mutants that are deleterious toC3b or C4b binding (52). Whereas such mutations in SCRs 15 and16 are localized on one face of the molecule, those in SCR 17 arelocated on the opposite face of the molecule and require a swivelabout the SCR 16–17 linker segment to align with what is thepresumed ligand contact face contributed by SCRs 15 and 16. Ifthe initial contact of C4b is with the first two domains of either site1 or 2, then the conformational component to the kinetic analysismay reflect the swivel required to enable the contact with residuesin the third SCR domain contributing to the binding site.

The debate about the possible association of SLE with complete,or even partial, C4A deficiency states is at present not settled. Ithas been noted (11) that most of the early C4 population studiesrelied on C4 protein allotyping in the context of a two-locus perallelic chromosome genetic model. In these studies, the presenceof a null allele was derived from the comparison of the relativeprotein levels of the two isotypes. Recent genetic studies haverevealed a far more complex picture for the organization of the C4gene. In the current model, the C4 gene is part of a module con-taining the RP-C4-CYP21-TNX genes (RCCX module) that can bepresent in single, double, triple, or even quadruple copies on eachallelic chromosome. In turn, each C4 gene locus may encode eitherC4A or C4B protein (11). Therefore, precise C4 genotyping, in-cluding module-specific RFLP and PCR analysis to determine the

1678 ANALYSIS OF CR1 BINDING TO C4Ab AND C4Bb

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

exact number of C4 genes (53), is necessary in order not to mis-interpret the overdosage of one isotype as a partial deficiency ofthe other isotype. Nevertheless, even recent studies that have in-cluded the appropriate genotyping at the DNA level continue toshow strikingly conflicting results on the association of C4A de-ficiency states and SLE (28, 37). Although we cannot resolve thiscontroversy, if the selective lack of C4A does indeed contribute tothe development of SLE in at least some ethnic populations, ourresults suggest that it is not due to differences in the intrinsic bind-ing affinity between C4Ab and C4Bb for CR1. This being said, ifour in vitro observations on the greater propensity of C4Ab to formdisulfide-linked dimers, and its unique ability to form higher mo-lecular mass oligomers via its isotypic residue C1102, were to alsooccur in vivo, the clustering effect of the ligand may confer ahigher avidity interaction with CR1 for C4Ab-opsonized targets. Inkeeping with an earlier suggestion (25), such a higher functionalaffinity for CR1 of C4Ab-opsonized targets could enhance the ef-fect due to the inherently higher covalent binding efficiency toamino group-rich targets, such as immune complexes, of C4Abrelative to C4Bb. These combined effects may therefore confer anadvantage to having high levels of C4A for the disposal of immunecomplexes, and perhaps apoptotic bodies, a disposal process that isthought to be instrumental in the etiology of SLE (26).

AcknowledgmentsWe are indebted to Dr. John R. Glover of the Department of Biochemistry,University of Toronto, for generous access to his BIAcore X instrument.

References1. Lay, W. H., and V. Nussenzweig. 1968. Receptors for complement of leukocytes.

J. Exp. Med. 128:991.2. Cooper, N. R. 1969. Immune adherence by the fourth component of complement.

Science 165:396.3. Klickstein, L. B., W. W. Wong, J. A. Smith, J. H. Weis, J. G. Wilson, and

D. T. Fearon. 1987. Human C3b/C4b receptor (CR1): demonstration of longhomologous repeating domains that are composed of the short consensus repeatscharacteristic of C3/C4 binding proteins. J. Exp. Med. 165:1095.

4. Hourcade, D., D. R. Miesner, J. P. Atkinson, and V. M. Holers. 1988. Identifi-cation of an alternative polyadenylation site in the human C3b/C4b receptor(complement receptor type 1) transcriptional unit and prediction of a secretedform of complement receptor type 1. J. Exp. Med. 168:1255.

5. Klickstein, L. B., T. J. Bartow, V. Miletic, L. D. Rabson, J. A. Smith, andD. T. Fearon. 1988. Identification of distinct C3b and C4b recognition sites in thehuman C3b/C4b receptor (CR1, CD35) by deletion mutagenesis. J. Exp. Med.168:1699.

6. Krych, M., D. Hourcade, and J. P. Atkinson. 1991. Sites within the complementC3b/C4b receptor important for the specificity of ligand binding. Proc. Natl.Acad. Sci. USA 88:4353.

7. Kalli, K. R., P. H. Hsu, T. J. Bartow, J. M. Ahearn, A. K. Matsumoto,L. B. Klickstein, and D. T. Fearon. 1991. Mapping of the C3b-binding site ofCR1 and construction of a (CR1)2-F(ab )2 chimeric complement inhibitor.J. Exp. Med. 174:1451.

8. Krych, M., R. Hauhart, and J. P. Atkinson. 1998. Structure-function analysis ofthe active sites of complement receptor type 1. J. Biol. Chem. 273:8623.

9. Reilly, B. D., S. C. Makrides, P. J. Ford, H. C. Marsh, Jr., and C. Mold. 1994.Quantitative analysis of C4b dimer binding to distinct sites on the C3b/C4b re-ceptor (CR1). J. Biol. Chem. 269:7696.

10. Krych, M., L. Clemenza, D. Howdeshell, R. Hauhart, D. Hourcade, andJ. P. Atkinson. 1994. Analysis of the functional domains of complement receptortype 1 (C3b/C4b receptor; CD35) by substitution mutagenesis. J. Biol. Chem.269:13273.

11. Blanchong, C. A., E. K. Chung, K. L. Rupert, Y. Yang, Z. Yang, B. Zhou,J. M. Moulds, and C. Y. Yu. 2001. Genetic, structural and functional diversitiesof human complement components C4A and C4B and their mouse homologues,Slp and C4. Int. Immunopharmacol. 1:365.

12. Yu, C. Y., R. D. Campbell, and R. R. Porter. 1988. A structural model for thelocation of the Rodgers and the Chido antigenic determinants and their correla-tion with the human complement component C4A/C4B isotypes. Immunogenetics27:399.

13. Isenman, D. E., and J. R. Young. 1984. The molecular basis for the difference inimmune hemolysis activity of the Chido and Rodgers isotypes of human com-plement component C4. J. Immunol. 132:3019.

14. Law, S. K., A. W. Dodds, and R. R. Porter. 1984. A comparison of the propertiesof two classes, C4A and C4B, of the human complement component C4. EMBOJ. 3:1819.

15. Carroll, M. C., D. M. Fathallah, L. Bergamaschini, E. M. Alicot, andD. E. Isenman. 1990. Substitution of a single amino acid (aspartic acid for his-

tidine) converts the functional activity of human complement C4B to C4A. Proc.Natl. Acad. Sci. USA 87:6868.

16. Sepp, A., A. W. Dodds, M. J. Anderson, R. D. Campbell, A. C. Willis, andS. K. Law. 1993. Covalent binding properties of the human complement proteinC4 and hydrolysis rate of the internal thioester upon activation. Protein Sci.2:706.

17. Dodds, A. W., X. D. Ren, A. C. Willis, and S. K. Law. 1996. The reactionmechanism of the internal thioester in the human complement component C4.Nature 379:177.

18. Law, S. K., and A. W. Dodds. 1997. The internal thioester and the covalentbinding properties of the complement proteins C3 and C4. Protein Sci. 6:263.

19. Kishore, N., D. Shah, V. M. Skanes, and R. P. Levine. 1988. The fluid-phasebinding of human C4 and its genetic variants, C4A3 and C4B1, to immunoglobu-lins. Mol. Immunol. 25:811.

20. Schifferli, J. A., and J. P. Paccaud. 1989. Two isotypes of human C4, C4A andC4B have different structure and function. Complement Inflamm. 6:19.

21. Paul, L., V. M. Skanes, J. Mayden, and R. P. Levine. 1988. C4-mediated inhi-bition of immune precipitation and differences in inhibitory action of geneticvariants, C4A3 and C4B1. Complement 5:110.

22. Ebanks, R. O., A. S. Jaikaran, M. C. Carroll, M. J. Anderson, R. D. Campbell, andD. E. Isenman. 1992. A single arginine to tryptophan interchange at �-chainresidue 458 of human complement component C4 accounts for the defect inclassical pathway C5 convertase activity of allotype C4A6: implications for thelocation of a C5 binding site in C4. J. Immunol. 148:2803.

23. Gatenby, P. A., J. E. Barbosa, and P. J. Lachmann. 1990. Differences betweenC4A and C4B in the handling of immune complexes: the enhancement of CR1binding is more important than the inhibition of immunoprecipitation. Clin. Exp.Immunol. 79:158.

24. Gibb, A. L., A. M. Freeman, R. A. Smith, S. Edmonds, and E. Sim. 1993. Theinteraction of soluble human complement receptor type 1 (sCR1, BRL55730)with human complement component C4. Biochim. Biophys. Acta 1180:313.

25. Reilly, B. D., and C. Mold. 1997. Quantitative analysis of C4Ab and C4Bbbinding to the C3b/C4b receptor (CR1, CD35). Clin. Exp. Immunol. 110:310.

26. Pickering, M. C., M. Botto, P. R. Taylor, P. J. Lachmann, and M. J. Walport.2001. Systemic lupus erythematosus, complement deficiency, and apoptosis. Adv.Immunol. 76:227.

27. Schur, P. H., D. Marcus-Bagley, Z. Awdeh, E. J. Yunis, and C. A. Alper. 1990.The effect of ethnicity on major histocompatibility complex complement allo-types and extended haplotypes in patients with systemic lupus erythematosus.Arthritis Rheum. 33:985.

28. Dragon-Durey, M. A., N. Rougier, J. P. Clauvel, S. Caillat-Zucman, P. Remy,L. Guillevin, F. Liote, J. Blouin, F. Ariey, B. U. Lambert, et al. 2001. Lack ofevidence of a specific role for C4A gene deficiency in determining disease sus-ceptibility among C4-deficient patients with systemic lupus erythematosus (SLE).Clin. Exp. Immunol. 123:133.

29. Atkinson, J. P. 1989. Complement deficiency: predisposing factor to autoimmunesyndromes. Clin. Exp. Rheumatol. 7(Suppl. 3):S95.

30. Kemp, M. E., J. P. Atkinson, V. M. Skanes, R. P. Levine, and D. D. Chaplin.1987. Deletion of C4A genes in patients with systemic lupus erythematosus.Arthritis Rheum. 30:1015.

31. Howard, P. F., M. C. Hochberg, W. B. Bias, F. C. Arnett, Jr., and R. H. McLean.1986. Relationship between C4 null genes, HLA-D region antigens, and geneticsusceptibility to systemic lupus erythematosus in Caucasian and black Ameri-cans. Am. J. Med. 81:187.

32. Hartung, K., M. P. Baur, R. Coldewey, M. Fricke, J. R. Kalden, H. J. Lakomek,H. H. Peter, D. Schendel, P. M. Schneider, S. A. Seuchter, et al. 1992. Majorhistocompatibility complex haplotypes and complement C4 alleles in systemiclupus erythematosus: results of a multicenter study. J. Clin. Invest. 90:1346.

33. Olsen, M. L., R. Goldstein, F. C. Arnett, M. Duvic, M. Pollack, andJ. D. Reveille. 1989. C4A gene deletion and HLA associations in black Ameri-cans with systemic lupus erythematosus. Immunogenetics 30:27.

34. Dunckley, H., P. A. Gatenby, B. Hawkins, S. Naito, and S. W. Serjeantson. 1987.Deficiency of C4A is a genetic determinant of systemic lupus erythematosus inthree ethnic groups. J. Immunogenet. 14:209.

35. Fielder, A. H., M. J. Walport, J. R. Batchelor, R. I. Rynes, C. M. Black,I. A. Dodi, and G. R. Hughes. 1983. Family study of the major histocompatibilitycomplex in patients with systemic lupus erythematosus: importance of null allelesof C4A and C4B in determining disease susceptibility. Br. Med. J. (Clin. Res.Ed.) 286:425.

36. Kristjansdottir, H., K. Bjarnadottir, I. B. Hjalmarsdottir, G. Grondal, A. Arnason,and K. Steinsson. 2000. A study of C4AQ0 and MHC haplotypes in Icelandicmulticase families with systemic lupus erythematosus. J. Rheumatol. 27:2590.

37. Man, X. Y., H. R. Luo, X. P. Li, Y. G. Yao, C. Z. Mao, and Y. P. Zhang. 2003.Polymerase chain reaction based C4AQ0 and C4BQ0 genotyping: associationwith systemic lupus erythematosus in southwest Han Chinese. Ann. Rheum. Dis.62:71.

38. Prodeus, A. P., S. Goerg, L. M. Shen, O. O. Pozdnyakova, L. Chu, E. M. Alicot,C. C. Goodnow, and M. C. Carroll. 1998. A critical role for complement inmaintenance of self-tolerance. Immunity 9:721.

39. Chen, Z., S. B. Koralov, and G. Kelsoe. 2000. Complement C4 inhibits systemicautoimmunity through a mechanism independent of complement receptors CR1and CR2. J. Exp. Med. 192:1339.

40. Van den Elsen, J. M., A. Martin, V. Wong, L. Clemenza, D. R. Rose, andD. E. Isenman. 2002. X-ray crystal structure of the C4d fragment of humancomplement component C4. J. Mol. Biol. 322:1103.

1679The Journal of Immunology

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

41. Tack, B. F., J. Janatova, M. L. Thomas, R. A. Harrison, and C. H. Hammer. 1981.The third, fourth, and fifth components of human complement: isolation andbiochemical properties. Methods Enzymol. 80:64.

42. Janatova, J., and B. F. Tack. 1981. Fourth component of human complement:studies of an amine-sensitive site comprised of a thiol component. Biochemistry20:2394.

43. Hessing, M., J. Paardekooper, and C. E. Hack. 1993. Separation of different formsof the fourth component of human complement by fast protein liquid chroma-tography. J. Immunol. Methods 157:39.

44. Nagasawa, S., C. Ichihara, and R. M. Stroud. 1980. Cleavage of C4b by C3binactivator: production of a nicked form of C4b, C4b , as an intermediate cleav-age product of C4b by C3b inactivator. J. Immunol. 125:578.

45. Nagasawa, S., and R. M. Stroud. 1980. Purification and characterization of amacromolecular weight cofactor for C3b-inactivator, C4bC3bINA-cofactor, ofhuman plasma. Mol. Immunol. 17:1365.

46. Feldman, H. A. 1972. Mathematical theory of complex ligand-binding systems ofequilibrium: some methods for parameter fitting. Anal. Biochem. 48:317.

47. Weisman, H. F., T. Bartow, M. K. Leppo, H. C. Marsh, Jr., G. R. Carson,M. F. Concino, M. P. Boyle, K. H. Roux, M. L. Weisfeldt, and D. T. Fearon.1990. Soluble human complement receptor type 1: in vivo inhibitor of comple-ment suppressing post-ischemic myocardial inflammation and necrosis. Science249:146.

48. Isenman, D. E., and J. R. Young. 1986. Covalent binding properties of the C4Aand C4B isotypes of the fourth component of human complement on severalC1-bearing cell surfaces. J. Immunol. 136:2542.

49. Paccaud, J. P., J. L. Carpentier, and J. A. Schifferli. 1990. Difference in theclustering of complement receptor type 1 (CR1) on polymorphonuclear leuko-cytes and erythrocytes: effect on immune adherence. Eur. J. Immunol. 20:283.

50. Becherer, J. D., and J. D. Lambris. 1988. Identification of the C3b receptor-binding domain in third component of complement. J. Biol. Chem. 263:14586.

51. Jokiranta, T. S., J. Westin, U. R. Nilsson, B. Nilsson, J. Hellwage, S. Lofas,D. L. Gordon, K. N. Ekdahl, and S. Meri. 2001. Complement C3b interactionsstudied with surface plasmon resonance technique. Int. Immunopharmacol.1:495.

52. Smith, B. O., R. L. Mallin, M. Krych-Goldberg, X. Wang, R. E. Hauhart,K. Bromek, D. Uhrin, J. P. Atkinson, and P. N. Barlow. 2002. Structure of theC3b binding site of CR1 (CD35), the immune adherence receptor. Cell 108:769.

53. Blanchong, C. A., B. Zhou, K. L. Rupert, E. K. Chung, K. N. Jones, J. F. Sotos,W. B. Zipf, R. M. Rennebohm, and C. Yung Yu. 2000. Deficiencies of humancomplement component C4A and C4B and heterozygosity in length variants ofRP-C4-CYP21-TNX (RCCX) modules in Caucasians: the load of RCCX geneticdiversity on major histocompatibility complex-associated disease. J. Exp. Med.191:2183.

1680 ANALYSIS OF CR1 BINDING TO C4Ab AND C4Bb

by guest on February 12, 2018http://w

ww

.jimm

unol.org/D

ownloaded from